On Thu, Jul 1, 2010 at 8:41 PM, Don Dutile <ddutile@xxxxxxxxxx> wrote:
> stefano@xxxxxxxxxxxxxx wrote:
>> From: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
>>
>> Use xen_vcpuop_clockevent instead of hpet and APIC timers as main
>> clockevent device on all vcpus, use the xen wallclock time as wallclock
>> instead of rtc and use xen_clocksource as clocksource.
>> The pv clock algorithm needs to work correctly for the xen_clocksource
>> and xen wallclock to be usable, only modern Xen versions offer a
>> reliable pv clock in HVM guests (XENFEAT_hvm_safe_pvclock).
>>
>> Using the hpet as clocksource means a VMEXIT every time we read/write to
>> the hpet mmio addresses, pvclock give us a better rating without
>> VMEXITs. Same goes for the xen wallclock and xen_vcpuop_clockevent
>>
>> Signed-off-by: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
>> ---
>> arch/x86/xen/enlighten.c | 14 +--------
>> arch/x86/xen/suspend.c | 4 ++
>> arch/x86/xen/time.c | 59
>> ++++++++++++++++++++++++++++++++++---
>> arch/x86/xen/xen-ops.h | 7 +---
>> include/xen/interface/features.h | 3 ++
>> 5 files changed, 65 insertions(+), 22 deletions(-)
>>
>> diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
>> index 90bac21..bcd98a9 100644
>> --- a/arch/x86/xen/enlighten.c
>> +++ b/arch/x86/xen/enlighten.c
>> @@ -933,10 +933,6 @@ static const struct pv_init_ops xen_init_ops __initdata
>> = {
>> .patch = xen_patch,
>> };
>>
>> -static const struct pv_time_ops xen_time_ops __initdata = {
>> - .sched_clock = xen_sched_clock,
>> -};
>> -
>> static const struct pv_cpu_ops xen_cpu_ops __initdata = {
>> .cpuid = xen_cpuid,
>>
>> @@ -1074,7 +1070,6 @@ asmlinkage void __init xen_start_kernel(void)
>> /* Install Xen paravirt ops */
>> pv_info = xen_info;
>> pv_init_ops = xen_init_ops;
>> - pv_time_ops = xen_time_ops;
>> pv_cpu_ops = xen_cpu_ops;
>> pv_apic_ops = xen_apic_ops;
>>
>> @@ -1082,13 +1077,7 @@ asmlinkage void __init xen_start_kernel(void)
>> x86_init.oem.arch_setup = xen_arch_setup;
>> x86_init.oem.banner = xen_banner;
>>
>> - x86_init.timers.timer_init = xen_time_init;
>> - x86_init.timers.setup_percpu_clockev = x86_init_noop;
>> - x86_cpuinit.setup_percpu_clockev = x86_init_noop;
>> -
>> - x86_platform.calibrate_tsc = xen_tsc_khz;
>> - x86_platform.get_wallclock = xen_get_wallclock;
>> - x86_platform.set_wallclock = xen_set_wallclock;
>> + xen_init_time_ops();
>>
>> /*
>> * Set up some pagetable state before starting to set any ptes.
>> @@ -1330,4 +1319,5 @@ void __init xen_hvm_guest_init(void)
>> register_cpu_notifier(&xen_hvm_cpu_notifier);
>> have_vcpu_info_placement = 0;
>> x86_init.irqs.intr_init = xen_init_IRQ;
>> + xen_hvm_init_time_ops();
>> }
>> diff --git a/arch/x86/xen/suspend.c b/arch/x86/xen/suspend.c
>> index 6ff9665..0774c67 100644
>> --- a/arch/x86/xen/suspend.c
>> +++ b/arch/x86/xen/suspend.c
>> @@ -28,8 +28,12 @@ void xen_pre_suspend(void)
>>
>> void xen_hvm_post_suspend(int suspend_cancelled)
>> {
>> + int cpu;
>> xen_hvm_init_shared_info();
>> xen_callback_vector();
>> + for_each_online_cpu(cpu) {
>> + xen_setup_runstate_info(cpu);
>> + }
>> }
>>
>
> As I found out on older xen (non 'modern Xen version' ;-) )
> w/o XENFEAT_hvm_safe_pvclock, the above patch should look like:
>
> if (xen_feature(XENFEAT_hvm_safe_pvclock)) {
> int cpu;
> for_each_online_cpu(cpu) {
> xen_setup_runstate_info(cpu);
> }
> }
>
Thank you for testing my series and for the patch, I'll include it in
the next version.
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home